Prey Capture Behavior in the Blue-Tongued Skink, Tiliqua Scincoides

Prey Capture Behavior in the Blue-tongued Skink, Tiliqua scincoides

TAMARA 1. SMITH,1 KENNETH V, KARDONG,I,2 AND VINCENT 1. BELS3

'Department of Zoology, Washington State University, Pullman, Washington 99164-4236, USA and E-mail: [email protected] 3Institute of Agriculture, Agronomic Center of Applied Research in Hainut, rue paul PastUT, 11, B-7800 Ath, Belgium

ABSTRACT.-Squamate prey capture evolved in two general directions; one toward an emphasis upon lingual prehension and the other toward an emphasis upon jaw prehension. In basal squamates (lguania), lingual prehension characterizes prey capture. All other squamates (Scleroglossa) tend to use their jaws for prey prehension and the role of the tongue as a prehensile organ is reduced. However, within some scierogJossan lizards, lingual and jaw modes of prehension are present Selection of a distinct prehension mode during a feeding bout in these lizards has been hypothesized to be related to prey size. To test for the presence of lingual prehension and correlation with prey size, we examined feeding behavior in the blue-tongued skink, Tiliqua scincoides using two prey types (mealworm and cricket>. We confirmed that this skink uses both lingual and jaw modes of prehension with accompanying characteristic jaw kinematic profiles. With crickets, only jaw prehension was exhibited, but both modes were used when feeding on equivalently sized prey, mealworms. Consequently, prehension mode is not exclusively elicited by prey size. We, therefore, hypoth esize that selection of prehension modes, lingual or jaws, in these basal scleroglossans also includes prox imate factors related to prey behavior.

The tongue of iguanian lizards is used in prey ceptions. In herbivorous iguanians, the tongue capture (Throckmorton, 1976; Smith, 1984; may be used to draw the leaf or food to the Schwenk and Throckmorton, 1989; Bell, 1990; mouth and crop it with the jaws. But occasion Bels and Goosse, 1990; Kraklau, 1991; Delheusy ally the jaws are used to grasp and tear vege and Bels, 1992), an ancestral state for Squamata tation. In the sister group, Sphenodon, jaw pre (Schwenk, 1988; Schwenk and Throckmorton, hension is used to secure large prey (Gorniak et 1989). But within the more derived scleroglos aI., 1982; Schwenk and Throckmorton, 1989). sans, there is a tendency for the tongue to be However, SphenodOl1 and all iguanians use lin come specialized for chemoreception (Burg gual prehension for small prey (Schwenk and hardt, 1970; Schwenk, 1988, 1993), and its role Throckmorton, 1989). in prey capture lost as the jaws assume the ma Exceptions to the general pattern of prehen jor role in prehension. Each feeding mode, lin sion by means of jaws also exist for scleroglos gual and jaw prehension, is accompanied by a sans. Trachydosaurus rugosus (Scincidae), feeding characteristic kinematic profile (Fig. 1) (Bramble on snails, initially contact the prey with the and Wake, 1985; Schwenk and Throckmorton, tongue which drew the snail a short distance 1989; Goosse and Bels, 1992; Urbani and Bels, across the substrate until the jaws secured the 1995; Bels and Kardong, unpubl. data). For ig prey (Gans et aI., 1985). The snail is never lifted uanians, the kinematic pattern accompanying by the tongue and carried into the mouth, a be tongue prehension consists of four stages. Jaw havior interpreted as kinematically distinct from opening begins with a slow open I (SOl), fol that recorded in iguanians (Schwenk and lowed by a slower, slow open II stage (SOIl), Throckmorton, 1989). In another scleroglossan, and ends with a fast open stage (FO) as the Zonosaurus laticaudatus (Corydiylidae), jaws were tongue projects from the mouth (Schwenk and used with large prey, but tongue use was used Throckmorton, 1989; Bels et aI., 1994). During with small prey; it was concluded that the jaw closing, the jaws rapidly close (FC) (Fig. 1A). kinematic profiles accompanying tongue use For some scleroglossans, the kinematic profile represented an ancestral prehension profile as accompanying jaw prehension consists of a fast observed in iguanian lizards (Urbani and Bels, open (FO) and fast close stage (FC), and the 1995). tongue does not protrude from the mouth (Fig. While jaw prehension is the sole mode of cap IB). ture in derived scleroglossans, even with very .~ In general, lingual prehension characterizes small prey (Smith, 1982, 1986; Schwenk and the iguanian clade, and jaw prehenSion charac Throckmorton, 1989), there is evidence that terizes the scleroglossan clade, but there are ex- tongue use is present in some groups of basal scleroglossan lizards and choice of feeding 2 Corresponding Author. mode is related to size of the prey (Urbani and LIZARD FEEDING 363

A B

.o·

SOl SOIl Fa Fe FO Fe

FIG. 1. Characteristic kinematic profile (modified from Bramble and Wake, 1985). (A.) The kinematic pattern accompanying lingual prehension consists of four stages. Jaw opening begins with a slow open I (501) followed by a slower, slow open II stage (SOD), and ends with a fast open stage (Fa) as the tongue projects from the mouth. During closing, the jaws rapidly close (FC). (B.) The kinematic profile accompanying jaw prehension consists of a fast open (Fa) and fast close stage (FC), and the tongue does not project from the mouth.

Bels, 1995). It has been hypothesized that when tive bred blue-tongued skinks, Tiliqua scincoides, lingual prehension is present in a species, a obtained from a commercial dealer. All had switch in prehension mode is mediated by the been in captivity over six months before feeding size ratio of predator to prey, a consequence of trials were initiated. Each lizard was isolated in changing surface area of the tongue relative to a terrarium (100cm X 50cm X 50cm) before the prey size (Bramble and Wake, 1985; filming. An incandescent bulb and two True Schwenk and Throckmorton, 1989; Urbani and Lite tubes provided the animal with a temper Bels, 1995). Consequently, the prehension mode ature of 21 C (night) and 29 C (d). The relative is largely determined by the wet adhesive prop humidity was maintained near 70%. Animals erties of the tongue relative to prey size, such were conditioned to feed on the cage floor in that, as the prey becomes larger the tongue loses front of a reference grid (10 X 10 mm). In film its effectiveness and the lizard must capture ing trials the prey consisted of equivalently prey using jaw prehension (Bramble and Wake, sized mealworm larvae (mean size: 23 :!:: 11 1985; Schwenk and ThrockulOrton, 1989). The mm) and crickets (mean size: 31 :!:: 8mm). significance of lingual prehension in some spe Feeding Trials.-For high-speed cinematogra cies of scleroglossans is disputed as whether it phy, four adult blue-tongued skinks, Tiliqua represents an isolated occurrence, or character scincoides (SVL: 27.11 :!:: 8.0cm) were filmed at izes these species as functional intermediates 100 frames/sec using Eastman Ektachrome between ancestral iguanian lizards and more high-speed 7250 tungsten 16 mm film, using a derived scleroglossans (Schwenk and Throck photosonic 1 PL camera under two 1000w tung morton, 1989). sten photoflood lights. At the initiation of a To characterize the relationships between mealwonn feeding trial, a preweighed and mea prey type and prehension mode, we exarn.ined sured mealwonn was placed in front of the liz feeding behavior in a scleroglosan species, the ard using long forceps. The camera was started blue-tongued skink, Tiliqua scincoides, which we and the ensuing predatory behavior filmed. For found to use both lingual and jaw prehension. a cricket feeding trial, one preweighed and mea We determined prehension mode when pre sured cricket was tied to the end of a line and sented with two prey types (mealwonns and placed on the surface in front of the lizard. A crickets) and examined whether or not the pre hension modes were correlated with prey size total of 44 mealwonn feeding trials were filmed. alone. Of these, 21 trials were of feeding in which the head of the skink did not tum and remained at MATERIALS AND METHODS right angles to the camera. These favorable Tiliqua scincoides is a scleroglossan lizard views were used for quantitative analysis. A to placed within the Scincidae (Estes et al., 1988). tal of 16 cricket feeding trials were filmed, of We adopted the squamate classi£cation present which seven trials were true lateral and the ed by Estes et al. (1988) because it is most com cricket was captured from the substrate. Frame plete and consistent with squamate monophyly. by frame quantitative analysis of these film se This phylogeny recognizes two major clades, quences was perfonned using a Pony-cam the iguania and the scleroglossa. 16mm cine projector. Lizards.-Data were gathered from four, cap Qualitative and Quantitative Analysis.-Kine 364 T. L. SMITH ET AL.

H time-to-close (TC), time from peak gape angle to jaw tips closed; and gape cycle time (GCT), time when jaws first begin to part to jaws tips closed. Maximum head rotation (MHR), the tilt ing of the snout downward as measured from the horizontal substrate, was analyzed for Th TP~ missed versus successful prey capture trials. L'I These data were not entirely independent, be cause we analyzed several feeding trials from FIG. 2. Digitized points, using natural and distinct markings on the lizard's body: UJ-Upper Jaw, LJ one animal. Due to the opportunistic design of Lower Jaw, G-Gape, TH-Throat, E-Eye, H-Head, the study, resulting in uneven sample sizes, any T-Tongue, and P-Prey. Reference bar 2 em. suitable nonparametric test resulted in" the loss of informative data. As a result, each trial was treated as an independent sample and two matic profiles of each feeding mode were deter tailed student t-tests were used to compare sta mined as described elsewhere (Delheusy and tistical variables between prehension modes. In Bels, 1992). From the lateral images, vertical y addition, a chi-square test was used for mea and horizontal x- coordinates of selected points suring the correlation between the mode of cap were recorded, using natural and distinct mark ture and the prey item. Of the four animals ings on the lizard's body (Fig. 2). Points at the used, no single individual accounted for more anterior tips of the upper and lower jaw (VI and than 30% of the data. ) LJ the comer of the mouth, gape (G), and the eye (E) were digitized. The head (H) was rep RESULTS resented by a point immediately posterior to the General Patterns.-The blue-tongued skink, Ti cervical region, the throat (TH) was digitized as liqua scincoides when presented with individual, natural marking at the middle of the throat for equivalently sized mealworms, used both jaw each individual lizard, and a point on the and lingual modes of prehension in capturing tongue (T) was selected and followed through the prey (Fig. 3). When presented with crickets, out the feeding episode as the portion of the only jaw prehension was used. The mode of pre tongue first leaving the buccal cavity. The prey hension, jaw or lingual, was accompanied by a (P), mealworm or cricket, was digitized at the characteristic kinematic pattern. anterior-most point on each individual. Project Approach and capture stages of a feeding tri ed images were digitized using a Summagraph al often consisted of several tries, lingual or ics MM 1201 digitizing tablet and analyzed us jaws, preceding successful prey capture. Cap ing Lotus 123 and Statmost (DataMost Corp) ture success was similar, 32.3% (21/65) and software packages. All x- and y- coordinates 34.4% (11/32), for mealworm versus cricket plotted as graphs were corrected to avoid the (Fig. 4A). In none of the 21 mealworm trials was effects of vertical rotation of the head. For each a missed lingual attempt followed immediately trial, the fixed grid, divided into 10 mm square by a jaw prehension attempt. However, in many units and placed immediately behind the lizard, (N = 15) mealworm trials, a missed jaw capture was used as the inertial reference frame. attempt was followed by successful lingual cap Statistical Variables.-The prey capture behav ture. While the tongue and jaws were used ior was scored as lingual if the foretongue equally during missed attempts, the tongue was moved anteriorly from the mouth and protrud used more frequently during a success (Fig. 4B). ed outside the buccal cavity. Capture was After a missed attempt(s), the lizard made be scored as jaw prehension if the tongue remained havioral modifications, changing head and body within the buccal cavity during prey capture. positioning. This included tilting the snout The following accompanying kinematic vari downward at a more direct angle to the prey, ables for capture cycles were scored for the two lifting the body at the shoulders, or head rota modes of prehension: Maximum gape angle tion about the long axis of the skull thereby pre (MGA), maximum gape angle, calculated from senting one side of the jaws to the prey. Often the points Vj, G, and Lj, during the prey capture these head rotations and postural changes were trial; time-to-peak gape (TPG) time when jaws used in combination (Fig. 5). A missed attempt, first begin to part to maximum gape angle; either lingual or jaw, was characterized by less

FIG. 3. Prey capture. Jaw prehenSion (left). Lingual prehenSion (right). Elapsed time shown in comer of frames. LIZARD FEEDING 365 366 T. L. SMITH ET AL.

A B

100 100

D TONGUE D MISS

• JAWS • CAPTURE 50

MISS CAPTURE MEALWORM CRlCKET MEALWORM FIG. 4. Prey Capture Success (percentage %). (A.) A feeding trial often consisted of several missed attempts, mealworm or cricket, preceding successful prey capture. (5.) While the tongue and jaws were used equaJJy during missed attempts on mealworms, the tongue was used more frequently during a success. head rotation than in a successful capture se als, 11 involved lingual prehension. With the quence (36.14 vs. 46.7 degrees) (Table 1). In presentation of an individual mealworm, the many of these missed attempts, the head tended lizard positioned its head close to and in front to be held horizontal to the substrate. This ap of the prey during the approach stage. In most, peared to produce an unfavorable orientation of 72.7% (8/11) the lower jaw made contact with the jaw tips to the prey. The lizard tended to the substrate followed by tongue protraction. either push the prey away with the forward The tongue protracted toward the mealworm moving lower jaw or it did not gain a firm grasp and curled so that the tongue tips curved ven on the prey between the jaw tips. trally and caudally, exposing the wider dorsal Statistical variables for capture cycles were surface of the tongue to the prey (Fig. 3). compared for the two modes of prehension, lin In the majority, 81.8% (9/11), of lingual pre gual versus jaw (Table 1). The time to peak gape hension trials, the lower jaw exhibited no for (0.25 vs. 0.15s, t = 3.87, P < 0.01) and gape cycle ward displacement while the tongue retracted. time (0.37 vs. 0.27s, t = 2.61, P < 0.05) were In the remaining tongue prehension trials, significantly faster during jaw prehension. Nei 18.2% (2/11), the rate of tongue retraction ther maximum gape angle (23.26 vs. 17.70 de equaled the rate of forward movement of the grees, t = 2.03, P > 0.05), or time-to-close (0.11 head. In these trials, jaw advancement was not vs. 0.13, t = -1.11, P > 0.05) were significantly impeded by contact with the substrate or con different between lingual and jaw modes of pre tact with substrate indirectly via tongue-pin hension. A chi-square test shows a significant ning. In a characteristic capture bout, the re association between prey type and mode of pre tracting tongue carried the prey well back into hension (X 2 = 14.3; df = 1; P < 0.01). the mouth behind the mid-region of the buccal Lingual Prehension.-Df the 21 mealworm tri- cavity.

A B

FIG. 5. After a missed attempt(s), the lizard made behavioral modifications, changing head and body po sitioning. Rotation included tilting the snout downward at a more direct angle to the prey and lifting the body at the shoulders (A) thereby presenting the jaws to the prey (B). LIZARD FEEDING 367

TABLE 1. Statistical variables during head rotation and comparison between the two modes of prehension, lingual versus jaw for capture cycles (student t-test * P < .05, ** P < .01).

Miss Capture Variables N Mean SE N Mean SE t-test Maximum Head Rotation (MHR) (Degrees) 22 36.14 2.059 13 46.70 3.436 P < 0.0020.... Lingual Jaw prehension prehension N Mean SE N Mean SE t-test Maximum Gape Angle (MGA) (Degrees) 9 23.2609 2.1569 8 17.7018 1.6788 P < 0.0770 Time to Peak Gape (TPG) (Seconds) 9' 0.2467 0.0163 8 0.1525 0.0185 P < 0.0061 ** Time of Close (TC) (Seconds) 9 0.1178 0.0097 8 0.1350 0.0073 P < 0.3052 Gape Cycle Time (GCT) (Seconds) 9 0.3689 0.0238 8 0.2657 0.0275 P < 0.0398"

The kinematic profile accompanying lingual gual feeding in Tiliqua is a reversal, i.e. second prehension resembled the general iguanian ki arily re-evolved from a jaw-feeding ancestor nematic profile consisting of four stages: a short within Scleroglossa. initial slow open (Sal) stage, followed by a Our results also differ from reports of tongue slower, slow open (SOIl), a fast open (Fa) stage, use in lizards with specialized diets. In Trachy and completed in a fast close (FC) stage (Figure dosaurus rugosus, the tongue is not used to lift 6A). In the two trials where there was no jaw the prey (snail) and transport it to the buccal contact with the substrate, a well-defmed SOIl cavity (Gans et aI., 1985; Schwenk and throck stage was missing. These two trials were ana morton, 1989). However, in some of our trials lyzed kinematically independently of the other with T scincoides, the tongue was presented in lingual prehension trials. a similar position (curled) as seen in iguanians Jaw Prehensiol1.-In many, 42.9% (9/21), meal and used to actually transport the mealworm worm trials and in all, 100% (16/16), cricket tri into the mouth. In addition, unlike the distinc als the jaws were used during capture. When tive kinematic profile of T rugosus, the kinematic using the jaws for capturing, the lizard com profile of tongue prehension in the blue monly lunged in the direction of the prey and, tongued skink was similar to the general feed as with tongue-use, the depressed lower jaw ing pattern (Bramble and Wake, 1985). made contact with the substrate. Initially, the Unlike iguanians feeding on small prey, prey was grasped just inside the tips of the jaws. which are obligate tongue-feeders, and the vast The tongue was employed during the subse majority of scleroglossans, which are obligate quent gape cycle to transport the prey intraor jaw-feeders, Tiliqua is capable of switching its ally (Fig. 3). prehension mode. The proximate basis for se The kinematic profile accompanying jaw pre lecting prehenSion mode in this skink species is hension resembled the generalized scleroglos consequently of special interest. On theoretical san profile: gape angle increased and decreased (Bramble and Wake, 1985) and empirical at a constant rate, with fast open (Fa) and fast grounds (Schwenk and Throckmorton, 1989; Ur close (FC) stages, respectively (Fig. 6B). bani and Bels, 1995), increasing prey size is hy pothesized to favor jaw prehension; small prey DISCUSSION lingual prehension. Certainly our data are con For the blue-tongued skink Tiliqua scincoides, sistent with this hypothesis. The jaws were used two modes of prehension, lingual and jaw, are exclusively on crickets (large) compared to present during independent feeding trials. Each mealworms (small). However, our data suggest mode is associated with a characteristic jaw ki that other proximate factors might be involved nematic profile, and these profiles are similar to as well. those of prehension reported in other lizard When presenting skinks with the same sized groups. Although lingual prehension in Tiliqua prey (mealworms), we expect that the same is kinematically similar to lingual prehension in prey capture mode would be used. This did not iquanians, the phylogenetic position of skinks happen. The lizards used both prehension within Scleroglossa strongly suggests that lin modes, not just lingual prehension. Although 368 T. L. SMITH ET AL.

30..,------,------r----,------,

Ul Q) 20 ~ Cl e.Q) w ..J C) «Z w 10 D« C)

0+------,-+-----,.-1-----+-..------' o .12 .24 .48 501 5011 FO Seconds

B 22,------,------_

17 Ul Q) ~ Cl e.Q) w 11 ..J C) «Z W «D 6 C)

O+---~--r--~-_,_+_~--..,._-~-_i o .12 .24 .36 .48 FO Fe Seconds FIG. 6. (A.) The kinematic profile accompanying lingual prehension displayed as the mean gape angle with error (standard error) bars. (B.) The kinematic profile accompanying jaw prehension displayed as the mean gape angle with error (standard error) bars. not quantified, we noted that the activity of a rapid means of prey capture Gaw) would there mealworm changed in several ways during the fore be likely when the prey is active and the approach of the lizard. The mealworm might be risk of escape great. This would suggest jaw motionless or moving. If moving it might be prehension with mealworms was induced by traveling away from the lizard or in one of the prey-activity, not prey size, per se. This would other possible directions (toward, across). Such also suggest why with a very active prey, crick "evasive" action by the mealworm increases the ets, jaw prehension was predominant. chance the lizard will lose the prey. We did de Thus the selection of feeding modes is likely termine that jaw prehension was significantly related to factors in addition to prey size faster, to peak gape, than lingual prehension. A (length). Other factors may include prey move LIZARD FEEDING 369 ment, orientation of prey presentation, height of GANS, c., F. DEVREE, AND D. CARRIER. 1985. Usage prey above the substrate, as well as factors re pattern of the complex masticatory muscles in the lated to size, such as the diameter silhouette of shingleback lizard, TraclnJdosaurus rugosus: A mod the prey and the prey's actual mass. el for muscle placement. Amer. J. Anat. 173:219 240. Acknawledgments.-Dur thanks go to D. E. GOOSSE, v., AND V. L. BELS. 1992. Kinematic and func Miller and P. A. Verrell who carefully read the tional analysis of feeding behaviour in Lacerta vir manuscript, and to Anthony Herrel for kindly idis (Reptilia: Lacertidae). Zool. ]b. Anat. 122:187 discussing with us his perspectives on skink 202. GORNIAK, G. c., H. I. ROSENBERG, AND C. GANS. 1982. feeding. In particular, we are grateful to K. Mastication in the Tuatara, Sphenodon punctatus Schwenk who reviewed the manuscript and (Reptilia: Rynchocephalia): Structure and activity generously shared with us his views on squa of the motor system. J. Morpho!. 171 :321-353. mate evolution. KRAKLAU, D. M. 1991. Kinematics of prey capture and chewing in the lizard Agama agama (Squamata: LITERATURE CITED Agamidae). J. Morpho!. 210:195-212. BELL, D. A. 1990. Kinematics of prey capture in the SCHWENK, K. 1988. Comparative morphology of the chameleon. Zool. ]b. Physio!. 94:247-260. Lepidosaur tongue and its relevance to Squamate BELS, V. L., M. CHARDON, AND K. V. KARDONG. 1994. phylogeny. In R. Estes and G. Pregill (eds.), Phy Biomechanics of the hyolingual system in Squa logenetic Relationships of the Lizard Families, pp. mata. In V. Bels, M. Chardon, P. Vanderwale, (eds.), 569-598. Stanford Univ. Press, Stanford. Advances in Comparative and Environmental ---. 1993. The evolution of chemoreception in Physiology, Vo!. 18. pp. 197-240. Springer-Verlag, squamate reptiles: a phylogenetic approach. Brain Berlin. Behav. Evol. 41:124-137. ---, AND V. GOOSSE. 1990. Comparative kinematic ---, AND G. S. THROCKMORTON. 1989. Functional analysis of prey capture in Anolis carolinenis (Igu and evolutionary morphology of lingual feeding in ania) and Lacerta viridis (Scleroglossa). J. Exp. Zool. squamate reptiles: phylogenetics and kinematics. J. 255:120-124. ZooI. (London) 219:153-175. BRAMBLE, D. M., AND D. B. WAKE. 1985. Feeding SMITH, K. K. 1982. An electromyographic study of the mechanisms of lower tetrapods. In M. Hilder function of the jaw adducting muscles in Varanus brand, D. M. Bramble, K. F. Liem, and D. B. Wake exanthemalicus (Varanidae). J. Morphol. 173:137 (eds.), Functional Vertebrate Morphology, pp. 230 158. 261. Harvard Univ. Press, Cambridge, Massachu ---. 1984. The use of the tongue and hyoid appa setts. ratus dUring feeding in lizards (Ctenosaura similis BURGHARDT, G. M. 1970. Chemical reception in rep and Tupinambis nigropunctatus). J. Zool. (London) tiles. In J. W. Johnston, Jr., D. G. Moulton, and A. 202:115-143. Turk (eds.), Communication by Chemical Signals, ---. 1986. Morphology and function of the tongue pp. 241-308. Appleton-Century-Crofts, New York. and hyoid apparatus in Varanus (Varanidae, Lac DELHEUSY, V., AND V. L. BELS. 1992. Kinematics of ertilia). J. Morphol. 187:261-287. feeding behaviour in Oplurus cuvieri (Reptilia: 19 THROCKMORTON, G. S. 1976. Oral food processing in uanidae). J. Exp. BioI. 170:155-186. two herbivorous lizards, Iguana iguana (Iguanidae) EsTES, R., ANO G. PREGILL (eds.). 1988. Phylogenetic and Uromastix aegyptus (Agamidae). J. Morphol. Relationships of the Lizard Families. Stanford 148:363-390. Univ. Press, Stanford, California. URBANI, ].-L. AND V. L. BELS. 1995. Feeding Behavior ---, K. DE QUEIROZ, AND J. GAUTHIER. 1988. Phy in two scleroglossan lizards: Lacerta viridis (Lacer logenetic relationships within Squamata. 1/1 R. Es tidae) and Zonosaurus laticaudatus (Cordylidae). J. tes and G. Pregill (eds.), Phylogenetic Relation Zool. (London) 236:265-290. ships of the Lizard Families, pp. 119-281. Stanford Univ. Press, Stanford, California. Accepted: 8 April 1999.